This paper presents a stochastic parking model for evaluating the effects of parking policies applied in urban areas. Previous parking models have used the network approach, which simultaneously simulates the parking and route choices, by using an extended network that considers some properly introduced parking and pedestrian links. The present model uses a non-network approach, which explicitly introduces a spearate choice submodel within the standard sequence of demand and assignment models. A full mathematical presentation of the proposed non-network approach is discussed. Definitions are given of parking alternatives, a random utility approach for parking choices, and cost functions associated with parking choices. A parking choice algorithm is specified in detail. The approach's data requirements are considered briefly, from the parking supply and demand and calibration viewpoints. Finally, a comparison is made between the non-network approach, as presented here, and an implementation of the network approach. In the network approach, the parking choice and route choice models must be calibrated jointly using the probit specification. In the non-network approach, the parking choice can be calibrated separately within an easier logit specification.
Abstract